Building a Prediction Model for Children's Social-emotional Development: A Community- Based Cohort in Shanghai, China

Social-emotional development (SED) delays in early childhood significantly increase risks of behavioral and mental health disorders. Identifying modifiable risk and protective factors is crucial for improving SEDs in children. This prospective cohort study aims to develop a predictive model for SED delays in 18-month-old children using data from 419 mother-child dyads in Shanghai, China. Twenty-one variables across maternal biomedical, psychological, socioeconomic, and infant domains were analyzed, with maternal anxiety and depression status tracked at five timepoints from pregnancy to 18 months postpartum. The SED status of offspring measured by the Ages & Stages Questionnaire: Social-Emotional. Extreme Gradient Boosting (XGBoost) feature selection identified five key predictors. The mean delivery age of the mothers was 32.2 years. All children were from full-term singleton births. Higher maternal age and maternal educational level, adequate sleep at 1-year-old may reduce SED delay risk, while maternal anxiety/depression status at 18 months postpartum may increase the risk. The model achieved excellent predictive performance (AUC = 0.93). Findings suggested that community-level screening of maternal mental health at 18 months postpartum and tailored measures (e.g., parental health education on child sleep patterns) to ensure adequate sleep for children could mitigate SED delays.